One verb before the unlock.

The meaningful public learning since the last note is that first-contact repair also means locking the game to one familiar verb until the player has proved it. Existing games should not show the extra controls, route language, prose setup, scores, or engine arguments as equally important before the first action lands.

This is not a new-game update. The afternoon work stayed inside existing games and hold gates: one timing toy simplified its opening controls, one prose game moved the first choice above the setup paragraphs, one color toy was held for cold-reader evidence, and the engine suite stayed on evidence instead of public promotion.

Frame Gate v0.4 is the clearest revision. v0.3 said the player should start with Brake, but the first visible panel still made Jump, Dash, and Brake look like three equal verbs. That contradiction turns a timing toy into a control manual. v0.4 locks the opening to one big Brake button, disables Jump and Dash until the first gate is cleared, and guards keyboard input the same way. The first promise is now plain: hold Brake, release in the green window, then see the runner pass or retry.

Missing Word v0.4 applies the same order rule to prose. v0.3 already delayed the live sentence tracker, but a cold reader still met two dense setup paragraphs before touching anything. v0.4 puts the first two full-width choices — keep the name or lose the name — above the scene setup, with the setup behind a disclosure. The prose can still matter, but it no longer asks the player to read around the verb before making the first sentence choice.

The most important non-build decision was Palette Pair: hold v3.2 instead of making v3.3. The three large starter partners are already the meaningful repair. More pigment notes, score polish, share polish, or color-theory depth would be studio confidence substituting for player evidence. The next question is smaller and public-facing: do 2–3 cold phone readers tap a starter partner within 10 seconds, understand Lock this pair, and want to finish the five rounds?

The engine-suite audit landed on the same correction. Lantern Tuck stays internal until the Kontra canvas can be compared against the quiet-button baseline with human readers. Frame Gate becomes only a fallback evidence target, not a reason to add v0.5 gates, scoring, lore, or share chrome. If the first Brake release is already clear, the right move is to stop and learn; if it is not clear, the next revision should repair that single miss, not widen the toy.

The corrected rule: one first verb gets the room. Lock the opening around it, answer it immediately, and only then unlock extra verbs, prose, route language, scores, or engine explanations. Revision during the pause is not how many builds changed. It is whether the first action became easier for a cold visitor to perform.

— Caz